Abstract
Data routinely collected through the United Network for Organ Sharing (UNOS) lack the detailed information on medical resource utilization and treatment costs required to accomplish for center-level comparisons of quality of care and cost for pediatric heart transplantation. We aimed to overcome this limitation by merging UNOS with the Pediatric Health Information Systems (PHIS) database, an administrative database containing inpatient, emergency department, ambulatory surgery, and observation unit information from over 40 not-for-profit, tertiary care pediatric hospitals. Utilizing a probabilistic match based on center, date of birth, recipient gender and transplant date within +/− 2 days, more than 90% of eligible UNOS patients (N=2264) were successfully merged to their corresponding PHIS records. Thirty-day and 1-year mortality rates observed for the merged cohort (3.2% and 9.0%, respectively) were comparable to those previously reported for pediatric heart transplants, as were the significant predictors of increased mortality. These results demonstrate that the established UNOS-PHIS cohort will provide a valid platform for subsequent research aimed at identifying center-level differences that could be exploited to optimize quality of care while minimizing cost across institutions.
Keywords: heart transplantation, hospitals, pediatric, mortality, research
INTRODUCTION
Pediatric cardiac transplantation is a complex, life-saving procedure that has been extensively studied in multiple publications. Many of the most impactful publications utilize the United Network for Organ Sharing (UNOS) database that includes extensive individual recipient and donor data. These data have been used in the evaluation of risk factors of mortality and graft failure, as well as a wide range of other clinical endpoints.1 In addition, recent work has focused on center level characteristics associated with mortality.
Despite the richness of the UNOS data, important data elements are not available for analyses. Specifically, the UNOS data include very limited data on medical care prior to transplant listing, do not include resource utilization data, and do not include medical cost data. However, these data elements are at the core of data from the Pediatric Health Information Systems (PHIS) database. We have previously used the PHIS database to study multiple clinical outcomes in pediatric oncology, including studies of induction mortality risk factors, center level variation in outcomes, and PHIS adjusted costs of therapy. Furthermore, we and others have merged PHIS data with other data sets. While UNOS data have been merged with other data sets,4,5 to our knowledge, no other investigators have published a PHIS-UNOS merge. Based on our prior merging work, we hypothesized that we would successfully merge at least 90% of merge eligible patients and that a merged UNOS-PHIS data set would allow estimation of 30-day and one-year mortality rates post-transplantation by individual center.
METHODS
UNOS is the private, non-profit organization that manages the nation's organ transplant system. UNOS utilizes a secure online database to manage Organ Procurement and Transplant network (OPTN) data pertaining to every organ donation and transplant event occurring nationwide since October 1987. Information collected in this context includes detailed pre- and post-transplant demographic and clinical information for donor and recipient such as primary diagnosis, patient severity status, follow up information regarding supportive care, mortality and graft status.
PHIS is an administrative database containing inpatient, emergency department, ambulatory surgery, and observation data from 43 pediatric hospitals nationwide. This accounts for approximately 85% of the free-standing children’s hospitals. Patients in the PHIS database are assigned a de-identified medical record number and can be tracked across multiple hospitalizations. Specifically, PHIS contains diagnosis and procedure codes (by ICD-9-CM codes), and billed resource utilization data (e.g., pharmaceutical, imaging, and laboratory resources) of inpatient hospital encounters from as early as January 1st, 1999.
OPTN/UNOS extracted a dataset consisting of all cardiac transplants during the period from January 1, 2004 through March 31, 2015. Potential cases were identified by querying PHIS for patients 21 years or younger with a procedure code for heart transplantation (37.51). Once this list of PHIS patients was identified, a probabilistic merge was conducted to match patients from OPTN/UNOS database with those identified in PHIS as having received a heart transplantation based on center, date of birth, recipient gender and transplant date within +/− 2 days.
Standard statistical methods were used to for univariate and bivariate analyses. Thirty day and one year mortality rates were computed for the merged population overall and by individual covariate (see Table 1 for listing). Multivariable log-binomial regression models were used to estimate crude and adjusted risk ratios (RR) and corresponding 95% confidence intervals (CI) comparing mortality rates by categories of each covariate. Covariates were included in the multivariable model if there was evidence of meaningful difference in mortality. In the case of highly correlated covariates, the strongest predictor was retained. Plots of hospital-specific 30-day and 1-year observed and model-based expected mortality rates were generated to assess variability in practice. All statistical analyses were performed using SAS (version 9.2, SAS Institute, Inc., Cary, NC).
Table 1.
Comparison of characteristics for patients between merged and unmerged patients
| Total UNOS Population |
UNOS Population at Non-PHIS Institutions |
UNOS Population at PHIS-contributing Institutions | ||||
|---|---|---|---|---|---|---|
| Total | Merged | Not Merged | p-valuea | |||
| N | 4253 | 1744 | 2509 | 2264 | 245 | |
| Characteristic | ||||||
| Sex | 0.485 | |||||
| Female | 1896 (44.6%) | 761 (43.6%) | 1135 (45.2%) | 1019 (45.0%) | 116 (47.4%) | |
| Male | 2357 (55.4%) | 983 (56.4%) | 1374 (54.8%) | 1245 (55.0%) | 129 (52.7%) | |
| Race/Ethnicity | 0.078 | |||||
| White | 2319 (54.5%) | 894 (51.3%) | 1423 (56.7%) | 1277 (56.4%) | 146 (59.6%) | |
| Black | 939 (22.1%) | 451 (25.9%) | 488 (19.4%) | 431 (19.0%) | 57 (23.3%) | |
| Asian | 170 (4.0%) | 64 (3.7%) | 106 (4.2%) | 96 (4.2%) | 10 (4.1%) | |
| Hispanic | 698 (16.4%) | 280 (16.1%) | 420 (16.7%) | 393 (17.4%) | 27 (11.0%) | |
| Other | 67 (1.6%) | 55 (3.1%) | 72 (2.9%) | 67 (3.0%) | 5 (2.0%) | |
| Insurance status | 0.236 | |||||
| Public | 2071 (48.7%) | 881 (50.5%) | 1190 (47.4%) | 1078 (47.6%) | 112 (47.1%) | |
| Private | 2048 (48.4%) | 808 (46.3%) | 1240 (49.4%) | 1121 (49.5%) | 119 (49.0%) | |
| Other | 120 (2.8%) | 44 (2.5%) | 76 (3.0%) | 64 (2.8%) | 12 (4.9%) | |
| Indication for Transplant | 0.265 | |||||
| Congenital | 1763 (41.5%) | 583 (33.4%) | 1178 (47.0%) | 1071 (47.3%) | 107 (43.7%) | |
| Mypopathy | 2485 (58.4%) | 1157 (66.3%) | 1328 (52.9%) | 1193 (52.7%) | 135 (55.1%) | |
| Age at transplant, years | 0.743 | |||||
| 0 to <1 | 1134 (26.7%) | 345 (19.8%) | 789 (31.4%) | 715 (31.6%) | 74 (30.2%) | |
| 1 to <5 | 792 (18.6%) | 269 (15.4%) | 522 (20.8%) | 471 (20.8%) | 51 (20.8%) | |
| 5 to <10 | 478 (11.2%) | 161 (9.2%) | 318 (12.7%) | 286 (12.6%) | 32 (13.1%) | |
| 10 to <15 | 767 (18.0%) | 301 (17.3%) | 468 (18.7%) | 427 (18.9%) | 41 (16.7%) | |
| 15 to <22 | 1082 (25.4%) | 668 (38.3%) | 412 (16.4%) | 365 (16.1%) | 47 (19.2%) | |
| Severity Measures | ||||||
| UNOS Status (at registration) | 0.245 | |||||
| 1a | 2728 (64.1%) | 997 (57.2%) | 1731 (69.0%) | 1549 (68.4%) | 182 (74.3%) | |
| 1b | 643 (15.1%) | 329 (18.9%) | 314 (12.5%) | 285 (12.6%) | 29 (11.8%) | |
| 2 | 816 (19.2%) | 387 (22.1%) | 429 (17.1%) | 396 (17.5%) | 33 (13.5%) | |
| UNOS Status (at transplant) | 0.143 | |||||
| 1a | 3445 (81.0%) | 1315 (75.4%) | 2130 (84.9%) | 1912 (84.5%) | 218 (89.0%) | |
| 1b | 512 (12.0%) | 246 (14.1%) | 266 (10.6%) | 247 (10.9%) | 19 (7.8%) | |
| 2 | 296 (7.0%) | 183 (10.5%) | 113 (4.5%) | 105 (4.6%) | 8 (3.3%) | |
| Waitlist time, days | 0.209 | |||||
| 0 to 16 | 1033 (24.3%) | 403 (23.1%) | 630 (25.1%) | 580 (25.6%) | 50 (20.4%) | |
| 17 to 45 | 1014 (23.8%) | 392 (22.5%) | 622 (24.8%) | 563 (24.9%) | 59 (24.1%) | |
| 46 to 103 | 1029 (24.2%) | 398 (22.8%) | 631 (25.1%) | 559 (24.7%) | 72 (29.4%) | |
| >103 | 1177 (27.7%) | 551 (31.6%) | 626 (25.0%) | 562 (24.8%) | 64 (26.1%) | |
| Mechanical Support - at transplant | 0.454 | |||||
| Yes | 2895 (68.1%) | 1174 (67.3%) | 1721 (68.6%) | 1520 (67.1%) | 201 (82.0%) | |
| No | 1358 (31.9%) | 570 (32.7%) | 788 (31.4%) | 744 (32.9%) | 44 (27.9%) | |
| Center Volume (based on mean number of transplants per year)b | 0.127 | |||||
| Low (<6.4) | 1321 (31.1%) | 1180 (67.7%) | 138 (5.5%) | 121 (5.3%) | 17 (6.9%) | |
| Medium (6.4–12.7) | 1731 (40.7%) | 400 (22.9%) | 1332 (53.1%) | 1192 (52.7%) | 140 (57.1%) | |
| High (>12.7) | 1201 (28.2%) | 164 (9.4) | 1039 (41.4%) | 951 (42.0%) | 88 (35.9%) | |
p-value for the comparison of merged and unmerged patients
Volumes of services, wage- and price-adjusted charges for each unit of service, and department-specific cost-to-charge ratios for each hospital were obtained from PHIS. Adjusted inpatient treatment costs were calculated by multiplying the adjusted charge by the hospital-specific cost-to-charge ratio for the relevant department then further adjusted to 2011 US dollars using the consumer price index. Adjusted total cost was calculated for each patient as the sum of the daily costs for a 30-day follow-up period beginning with the date of transplantation. Indirect costs and costs accrued at non-PHIS hospitals and in outpatient settings were not captured. The 30-day adjusted total cost were summarized as median and corresponding interquartile range for the population overall and by levels of covariates, as well as separately for patients who died within the first 30 days and those who did not.
The study protocol was approved by the CHOP Institutional Review Board. Additionally, the study plan was also reviewed and approved by UNOS prior to releasing the patient-identified OPTN data required for the merge with PHIS.
RESULTS
A total of 4,600 transplants to 4,494 patients were identified through OPTN/UNOS from January 1, 2004 through March 31, 2015. Of these, 2,509 were first transplants that occurred at PHIS institutions within the center-specific time periods when data were contributed to PHIS and thus eligible to be included in the merge. Of these, 2,264 (90.2%) were successfully merged.
Patients who were successfully identified in PHIS did not differ from patients who were not identified in PHIS in regard to gender, race, age at transplant, indication for transplant, waitlist time prior to transplant, insurance at transplant admission, UNOS status at registration or transplant, or life support requirements at transplant (Table 1). There was no obvious trend with respect to matching by year alone. With respect to institution, 18 of the 28 PHIS sites had match rates >90% (11 with rates >95%). However, there were a few sites with disproportionately lower overall match rates (<80%), and 50% of the unmatched first transplants were from three centers. Among those three institutions, the poor match rates were only during a subset of the study period in which they are documented by PHIS as not contributing all data elements. These institutions had a 100% match rate for the remainder of the study period. Excluding these sites from the merge during their period of incomplete submission results in a revised overall merge rate of 94.9%. Compared to the full UNOS population, patients included in the merged population were somewhat younger and slightly more likely to have a congenital indication for transplant.
Overall mortality rates within 30-days and one year were 3.2% and 9.0%, respectively. Risk factors for death within the two follow-up periods were similar. Infant age, congenital indication, shorter (i.e. less than 16 days) and longer waitlist times (i.e., greater than ~100 days), ECMO requirement at transplant, impaired hepatic function at transplant, and dialysis requirement at transplant were each significantly associated with a higher mortality risk based on bivariate analyses (Tables 2 and 3). Each of these factors, with the exception of shorter waitlist times, remained significantly predictive of mortality in multivariable analyses. Notably, center volume was not a significant predictor of mortality.
Table 2.
Crude and Adjusted Comparisons of 30-Day Post-Transplant Mortality
| 30 Day Mortality | |||
|---|---|---|---|
| Rate, n (%) | Crude RR (95% CI) | Adjusted RR (95% CI) | |
| Overall | 73 (3.2%) | N/A | |
| Characteristic | |||
| Sex | |||
| Female | 34 (3.3%) | 1 (ref) | |
| Male | 39 (3.1%) | 0.94 (0.60, 1.48) | |
| Race | |||
| White | 46 (3.6%) | 1 (ref) | |
| Black | 9 (2.1%) | 1.74 (0.76, 3.96) | |
| Asian | 6 (6.3%) | 0.58 (0.29, 1.17) | |
| Hispanic | 10 (2.5%) | 0.71 (0.36, 1.39) | |
| Other | 2 (3%) | 0.83 (0.21, 3.34) | |
| Insurance status | |||
| Public | 30 (2.8%) | 0.76 (0.48, 1.21) | |
| Private | 41 (3.7%) | 1 (ref) | |
| Other | 2 (3.1%) | 0.84 (0.21, 3.40) | |
| Age at transplant, year | |||
| 0 to <1 | 35 (4.9%) | 2.88 (1.35, 6.16) | 2.15 (0.96, 4.85) |
| 1 to <5 | 8 (1.7%) | 1 (ref) | 1 (ref) |
| 5 to <10 | 10 (3.5%) | 2.06 (0.82, 5.16) | 1.93 (0.73, 5.06) |
| 10 to <15 | 8 (1.9%) | 1.10 (0.42, 2.91) | 1.03 (0.38, 2.81) |
| 15 to <21 | 12 (3.3%) | 1.94 (0.80, 4.69) | 2.00 (0.81, 4.99) |
| Indication for Transplant | |||
| Congenital | 58 (5.5%) | 4.31 (2.46, 7.55) | 3.63 (1.90, 6.97) |
| Myopathy | 15 (1.3%) | 1 (ref) | 1 (ref) |
| Severity Measure | |||
| UNOS Status (at registration) | |||
| 1a | 52 (3.4%) | 0.78 (0.46, 1.34) | |
| 1b | 4 (1.4%) | 0.18 (0.11, 0.96) | |
| 2 | 17 (4.3%) | 1 (ref) | |
| UNOS Status (at transplant) | |||
| 1a | 67 (3.5%) | 1.23 (0.39, 3.83) | |
| 1b | 3 (1.2%) | 0.43 (0.09, 2.07) | |
| 2 | 3 (2.9%) | 1 (ref) | |
| Waitlist time, median (range) | |||
| 0–16 | 27 (4.7%) | 2.89 (1.37, 6.09) | 1.63 (0.74, 3.61) |
| 17–45 | 13 (2.3%) | 1.43 (0.62, 3.33) | 1.05 (0.43, 2.55) |
| 46–103 | 9 (1.6%) | 1 (ref) | 1 (ref) |
| >103 | 24 (4.3%) | 2.65 (1.24, 5.66) | 3.07 (1.38, 6.83) |
| Mechanical Support at Transplant | |||
| Yes | 62 (4.1%) | 2.76 (1.46, 5.21) | |
| No | 11 (1.5%) | 1 (ref) | |
| ECMO at transplant | |||
| Yes | 25 (18.4%) | 8.15 (5.19, 12.80) | 4.59 (2.52, 8.36) |
| No | 48 (2.3%) | 1 (ref) | 1 (ref) |
| Bilirubin | |||
| Quartile 1 | 3 (0.6%) | 1 (ref) | 1 (ref) |
| Quartile 2 | 7 (1.5%) | 2.56 (0.67, 9.86) | 2.23 (0.55, 9.06) |
| Quartile 3 | 20 (3.1%) | 5.33 (1.59, 17.85) | 4.97 (1.45, 17.04) |
| Quartile 4 | 42 (7.5%) | 12.85 (4.01, 41.19) | 8.85 (2.67, 29.33) |
| Dialysis prior to transplant | |||
| Yes | 10 (14.7%) | 5.30 (2.84, 9.90) | 2.68 (1.30, 5.50) |
| No | 60 (2.8%) | 1 (ref) | 1 (ref) |
| Gender Match | |||
| Yes | 38 (3.2%) | 1.03 (0.65, 1.61) | |
| No | 35 (3.3%) | 1 (ref) | |
| Weight Ratio | |||
| Quartile 1 | 15 (2.7%) | 1 (ref) | |
| Quartile 2 | 19 (3.3%) | 1.21 (0.62, 2.35) | |
| Quartile 3 | 15 (2.9%) | 1.06 (0.53, 2.12) | |
| Quartile 4 | 23 (4.0%) | 1.47 (0.77, 2.78) | |
| Center Volume | |||
| Low | 3 (2.5%) | 0.73 (0.23, 2.32) | |
| Medium | 33 (3.1%) | 0.92 (0.58, 1.44) | |
| High | 37 (3.4%) | 1 (ref) | |
Center volume was computed as the number of transplants in the previous 5 years before their transplant and categorized as low, medium and high based on cut points at the 25th and 75th percentiles.
Table 3.
Crude and Adjusted Comparisons of 1-Year Post-Transplant Mortality
| 1 Year Mortality | |||
|---|---|---|---|
| Rate, n (%) | Crude RR (95% CI) | Adjusted RR (95% CI) | |
| Overall | 203 (9.0%) | N/A | |
| Characteristic | |||
| Sex | |||
| Female | 92 (9.0%) | 1 (ref) | |
| Male | 111 (8.9%) | 0.99 (0.76, 1.29) | |
| Race | |||
| White | 121 (9.5%) | 1 (ref) | |
| Black | 36 (8.4%) | 0.88 (0.62, 1.26) | |
| Asian | 8 (8.3%) | 0.88 (0.44, 1.74) | |
| Hispanic | 31 (7.9%) | 0.83 (0.57, 1.21) | |
| Other | 7 (10.5%) | 1.10 (0.54, 2.27) | |
| Insurance status | |||
| Public | 98 (9.1%) | 0.99 (0.76, 1.29) | |
| Private | 103 (9.2%) | 1 (ref) | |
| Other | 2 (3.1%) | 0.33 (0.08, 1.33) | |
| Age at transplant, years | |||
| 0 to <1 | 102 (14.3%) | 1.87 (1.30, 2.68) | 1.51 (1.03, 2.20) |
| 1 to <5 | 36 (7.6%) | 1 (ref) | 1 (ref) |
| 5 to <10 | 14 (4.9%) | 0.64 (0.35, 1.17) | 0.62 (0.34, 1.14) |
| 10 to <15 | 24 (5.6%) | 0.74 (0.45, 1.21) | 0.80 (0.49, 1.32) |
| 15 to <21 | 27 (7.4%) | 0.97 (0.60, 1.56) | 1.10 (0.68, 1.77) |
| Indication for Transplant | |||
| Congenital | 156 (14.6%) | 3.70 (2.70, 5.07) | 3.09 (2.22, 4.31) |
| Myopathy | 47 (3.9%) | 1 (ref) | 1 (ref) |
| Severity Measure | |||
| UNOS Status (at registration) | |||
| 1a | 144 (9.3%) | 1.05 (0.74, 1.50) | |
| 1b | 21 (7.4%) | 0.83 (0.50, 1.40) | |
| 2 | 35 (8.8%) | 1 (ref) | |
| UNOS Status (at transplant) | |||
| 1a | 180 (9.4%) | 1.41 (0.68, 2.93) | |
| 1b | 16 (6.5%) | 0.97 (0.41, 2.29) | |
| 2 | 7 (6.7%) | 1 (ref) | |
| Waitlist time, median (range) | |||
| 0–16 | 69 (11.9%) | 1.71 (1.17, 2.48) | 1.35 (0.91, 2.00) |
| 17–45 | 40 (7.1%) | 1.02 (0.67, 1.56) | 0.96 (0.63, 1.48) |
| 46–103 | 39 (7.0%) | 1 (ref) | 1 (ref) |
| >103 | 55 (9.8%) | 1.40 (0.95, 2.08) | 1.67 (1.11, 2.49) |
| Mechanical Support at Transplant | |||
| Yes | 159 (10.5%) | 1.77 (1.28, 2.44) | |
| No | 44 (5.9%) | 1 (ref) | |
| ECMO at transplant | |||
| Yes | 46 (33.8%) | 4.58 (3.47, 6.06) | 2.81 (1.97, 4.02) |
| No | 157 (7.4%) | 1 (ref) | 1 (ref) |
| Bilirubin | |||
| Quartile 1 | 39 (7.6%) | 1 (ref) | 1 (ref) |
| Quartile 2 | 25 (5.4%) | 0.70 (0.43, 1.15) | 0.76 (0.47, 1.24) |
| Quartile 3 | 56 (8.8%) | 1.15 (0.78, 1.70) | 1.21 (0.81, 1.80) |
| Quartile 4 | 81 (14.5%) | 1.91 (1.33, 2.74) | 1.59 (1.09, 2.34) |
| Dialysis prior to transplant | |||
| Yes | 15 (22.1%) | 2.61 (1.63, 4.16) | 1.74 (1.04, 2.92) |
| No | 183 (8.5%) | 1 (ref) | 1 (ref) |
| Gender Match | |||
| Yes | 107 (9.0%) | 1.00 (0.77, 1.30) | |
| No | 96 (9.0%) | 1 (ref) | |
| Weight Ratio | |||
| Quartile 1 | 54 (9.8%) | 1 (ref) | |
| Quartile 2 | 56 (9.7%) | 0.99 (0.69, 1.41) | |
| Quartile 3 | 45 (8.1%) | 0.83 (0.57, 1.21) | |
| Quartile 4 | 48 (8.3%) | 0.85 (0.59, 1.23) | |
| Center Volume | |||
| Low | 9 (7.4%) | 0.76 (0.40, 1.47) | |
| Medium | 88 (8.3%) | 0.85 (0.65, 1.12) | |
| High | 106 (9.8%) | 1 (ref) | |
Center volume was computed as the number of transplants in the previous 5 years before their transplant and categorized as low, medium and high based on cut points at the 25th and 75th percentiles.
Across hospitals, there was moderate variability rates of 30-day (range: 0% to 9.4%; Figure 1, Panel A) and one-year mortality (range: 2.1% to 16.5%; Figure 1, Panel B). Overall, lower 30-day rates were not predictive of lower one-year mortality rates. Adjustment for individual level predictors of mortality explained more of center-level variability at one-year than at 30-days (Figure 1).
Figure 1.
Center level variability in 30-day (Panel A) and 1-year (Panel B) post-transplant mortality
Blue= Observed, Red= Expected
Total inpatient costs over the 30-day follow-up for the merged population overall and by level of covariates are summarized in Table 4. The median total inpatient cost from transplant was $254,530 (IQR: 176,642–346,451); $286,956 (IQR: 199,869–397,436) for congenital indications and $226,944 (IQR: 156,071–309,472) for myopathy.
Table 4.
Total PHIS Adjusted Inpatient Costs during the 30-Day Post-Transplant Follow-up Period
| Median | Interquartile Range | |
|---|---|---|
| Overall | $254,530 | (176,642–346,451) |
| Sex | ||
| Female | $254,495 | (174,860–340,160) |
| Male | $254,559 | (178,077–351,516) |
| Race | ||
| White | $255,614 | (178,827–351,814) |
| Black | $252,583 | (174,407–352,571) |
| Asian | $214,943 | (134,992–323,267) |
| Hispanic | $255,680 | (179,596–327,315) |
| Other | $276,918 | (196,092–382,758) |
| Insurance status | ||
| Public | $264,619 | (185,471–355,148) |
| Private | $250,192 | (174,525–342,051) |
| Other | $148,100 | (106,818–239,334) |
| Age at transplant, year | ||
| 0 to <1 | $279,029 | (196,891–384,076) |
| 1 to <5 | $241,612 | (169,296–326,437) |
| 5 to <10 | $259,758 | (168,702–362,048) |
| 10 to <15 | $237,009 | (170,295–321,420) |
| 15 to <21 | $221,861 | (157,323–325,865) |
| Indication for Transplant | ||
| Congenital | $286,956 | (199,869–397,436) |
| Myopathy | $226,944 | (156,071–309,472) |
| UNOS Status (at transplant) | ||
| 1a | $260,468 | (180,448–355,828) |
| 1b | $230,158 | (158,181–299,188) |
| 2 | $228,369 | (138,194–299,931) |
| Waitlist time, median (range) | ||
| 0–16 days | $253,464 | (174,819–361,967) |
| 17–45 days | $245,857 | (170,295–332,798) |
| 46–103 days | $266,941 | (180,144–353,127) |
| >103 days | $248,437 | (184,562–335,115) |
| Mechanical Support at Transplant | ||
| Yes | $264,859 | (182,044–364,859) |
| No | $236,701 | (163,043–311,866) |
| ECMO at transplant | ||
| Yes | $372,631 | (244,192–536,710) |
| No | $250,459 | (172,575–336,618) |
| Vital Status at 30 Days Post-Transplant | ||
| Alive | $252,555 | (175,181–341,029) |
| Deceased | $342,838 | (210,478–500,202) |
DISCUSSION
We present a successful merge between OPTN/UNOS and PHIS. As hypothesized, more than 90% of eligible patients were merged and no significant differences were observed in merged and unmerged patients at PHIS centers. Of note 18 out of the 28 total PHIS institutions had match rates greater than 90% with 11 institutions having match rates greater than 95%. Three institutions had disproportionally lower match rates for some portion of the study period. More than 50% of unmatched records were from three institutions which are documented by PHIS as not contributing all data components during some portion of the study period. The overall match rate was is increased to 95% when periods of incomplete data submission are excluded. In the absence of an individual-level review against the medical records, we are not able to determine the specific reasons for the non-matching of the remaining 5%. That said, the distributions of matched and unmatched patients were not meaningfully different. Additionally, our 90.2% overall match rate is consistent with that of previous PHIS linkages to other data sources performed by our group and others,6,7,8 including linkages to the Children’s Oncology Group cooperative group clinical trial databases (90.9%), National Surgical Quality Improvement Program-Pediatric (92.5%), and the Scientific Registry of Transplant Recipients (90.8%). Additionally, while there was a qualitative difference between the merged population and overall UNOS population with respect to the distributions of age and indication for transplant (correlated with age) this is likely due to PHIS institutions’ being restricted to a pediatric population. Thus, the established UNOS-PHIS cohort provides a valid representation of the experience of pediatric patients undergoing cardiac transplantation at free-standing tertiary pediatric institutions.
Overall 30-day and one-year mortality rates, 3.2% and 9.0%, respectively, are consistent with in-hospital mortality rates described by Law et al.9 Risk factors for mortality were also consistent with those identified in previous studies1,10 with the exception of center volume. This is likely due to center volume homogeneity among PHIS institutions. Specifically, high-volume centers were defined as performing an average of 12 transplants per year2. The median number of transplants performed by PHIS-contributing institutions is 10 transplants per year, placing most PHIS institutions in the medium-high volume classification. In addition, the effect of short waitlist time on mortality was mitigated when adjusting for severity of disease at presentation. This may be due to patients with higher acuity at presentation being prioritized for transplant, thus decreasing waitlist duration.
Adjustment for patient-level factors resolved more of the center-level variability at one year than at 30 days. This demonstrates that observed patient-level factors better explain variability in mortality at one year compared with 30 days. Factors other than individual characteristics such as institutional-level variability in post-transplant management may explain the discrepancy between observed and predicted 30-day mortality rates. Moreover, patients are more likely to receive inpatient care within 30 days of transplantation, and thus institutional factors are more likely to influence outcomes at 30 days.
In summary, this study represents the first successful merger of data from OPTN/UNOS with data from a hospital administrative healthcare resource. The combined data draw on the strengths of the two individual data sets and afford the opportunity to perform studies that are not possible with either database alone. For example, evaluations of inpatient treatment costs and resource utilization. Work is ongoing to evaluate center level variation in survival outcomes, resource utilization and costs with the goal of optimizing clinical care and minimizing hospital costs.
Acknowledgments
This research was supported by National Institutes of Health grant 1R01 CA133881 (Aplenc).
Footnotes
CONFLICTS OF INTEREST
None of the authors have any conflicts of interest to disclose.
AUTHORSHIP STATEMENTS
Concept/design: Getz, Aplenc; Data collection/management: Huang; Data analysis/interpretation: Getz, Li, Burstein, Rossano, Aplenc; Statistics: Li, Huang; Drafting article: Getz, He; Critical revision of article: Li, Huang, Burstein, Rossano, Aplenc; Approval of final article: All authors; Funding secured by: Aplenc.
References
- 1.Dipchand AI, Rossano JW, Edwards LB, Kucheryavaya AY, Benden C, Goldfarb S, Levvey BJ, Lund LH, Meiser B, Yusen RD, Stehlik J. The Registry of the International Society for Heart and Lung Transplantation: Eighteenth Official Pediatric Heart Transplantation Report--2015; Focus Theme: Early Graft Failure. J Heart Lung Transplant. 2015;34:1233–43. doi: 10.1016/j.healun.2015.08.002. [DOI] [PubMed] [Google Scholar]
- 2.Davies RR, Russo MJ, Hong KN, Mital S, Mosca RS, Quaegebeur JM, Chen JM. Increased short- and long-term mortality at low-volume pediatric heart transplant centers: should minimum standards be set? Annals of Surgery. 2011;253:393–401. doi: 10.1097/SLA.0b013e31820700cc. [DOI] [PubMed] [Google Scholar]
- 3.Lui C, Grimm JC, Magruder JT, Dungan SP, Spinner JA, Do N, Nelson KL, Cameron DE, Vricella LA, Jacobs ML. The Effect of Institutional Volume on Complications and Their Impact on Mortality After Pediatric Heart Transplantation. Ann Thoracic Surg. 2015;100:1423–31. doi: 10.1016/j.athoracsur.2015.06.016. [DOI] [PubMed] [Google Scholar]
- 4.Spector LG, Menk JS, Vinocur JM, Oster ME, Harvey BA, St Louis JD, Moller J, Kochilas LK. In-Hospital Vital Status and Heart Transplants After Intervention for Congenital Heart Disease in the Pediatric Cardiac Care Consortium: Completeness of Ascertainment Using the National Death Index and United Network for Organ Sharing Datasets. J Am Heart Assoc. 2016;5:e003783. doi: 10.1161/JAHA.116.003783. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 5.Wilk AR, Edwards LB, Edwards EB. The Effect of Augmenting OPTN Data With External Death Data on Calculating Patient Survival Rates After Organ Transplantation. Transplantation. 2017;101:836–43. doi: 10.1097/TP.0000000000001448. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 6.Aplenc R, Fisher BT, Huang YS, Li Y, Alonzo TA, Gerbing RB, Hall M, Bertoch D, Keren R, Seif AE, Sung L, Adamson PC, Gamis A. Merging of the National Cancer Institute-funded cooperative oncology group data with an administrative data source to develop a more effective platform for clinical trial analysis and comparative effectiveness research: a report from the Children's Oncology Group. Pharmacoepidemiol Drug Saf. 2012 May;21(Suppl 2):37–43. doi: 10.1002/pds.3241. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 7.Deans KJ, Cooper JN, Rangel SJ, Raval MV, Minneci PC, Moss RL. Enhancing NSQIP-Pediatric through integration with the Pediatric Health Information System. J Pediatr Surg. 2014 Jan;49(1):207–12. doi: 10.1016/j.jpedsurg.2013.09.058. Epub 2013 Oct 5. [DOI] [PubMed] [Google Scholar]
- 8.Godown J, Thurm C, Dodd DA, Soslow JH, Feingold B, Smith AH, Mettler BA, Thompson B, Hall M. A unique linkage of administrative and clinical registry databases to expand analytic possibilities in pediatric heart transplantation research. Am Heart J. 2017 Dec;194:9–15. doi: 10.1016/j.ahj.2017.08.014. Epub 2017 Aug 23. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 9.Law SP, Kim JJ, Decker JA, Price JF, Cabrera AG, Graves DE, Morales DLS, Heinle JS, Denfield SW, Dreyer WJ, Rossano JW. Hospital charges for pediatric heart transplant hospitalizations in the United States from 1997 to 2006. J Heart Lung Transplant. 2012;31:485–91. doi: 10.1016/j.healun.2011.12.008. [DOI] [PubMed] [Google Scholar]
- 10.Almond CS, Gauvreau K, Canter CE, Rajagopal SK, Piercey GE, Singh TP. A risk-prediction model for in-hospital mortality after heart transplantation in US children. Am J Transplant. 2012;12:1240–48. doi: 10.1111/j.1600-6143.2011.03932.x. [DOI] [PubMed] [Google Scholar]